include/wx/stringimpl.h

   1 ///////////////////////////////////////////////////////////////////////////////
   2 // Name:        wx/stringimpl.h
   3 // Purpose:     wxStringImpl class, implementation of wxString
   4 // Author:      Vadim Zeitlin
   5 // Modified by:
   6 // Created:     29/01/98
   7 // RCS-ID:      $Id$
   8 // Copyright:   (c) 1998 Vadim Zeitlin <zeitlin@dptmaths.ens-cachan.fr>
   9 // Licence:     wxWindows licence
  10 ///////////////////////////////////////////////////////////////////////////////
  11
  12 /*
  13     This header implements std::string-like string class, wxStringImpl, that is
  14     used by wxString to store the data. Alternatively, if wxUSE_STL=1,
  15     wxStringImpl is just a typedef to std:: string class.
  16 */
  17
  18 #ifndef _WX_WXSTRINGIMPL_H__
  19 #define _WX_WXSTRINGIMPL_H__
  20
  21 // ----------------------------------------------------------------------------
  22 // headers
  23 // ----------------------------------------------------------------------------
  24
  25 #include "wx/defs.h"        // everybody should include this
  26 #include "wx/wxchar.h"      // for wxChar, wxStrlen() etc.
  27
  28 #include <stdlib.h>
  29
  30 // ---------------------------------------------------------------------------
  31 // macros
  32 // ---------------------------------------------------------------------------
  33
  34 // implementation only
  35 #define   wxASSERT_VALID_INDEX(i) \
  36     wxASSERT_MSG( (size_t)(i) <= length(), _T("invalid index in wxString") )
  37
  38
  39 // ----------------------------------------------------------------------------
  40 // global data
  41 // ----------------------------------------------------------------------------
  42
  43 // global pointer to empty string
  44 extern WXDLLIMPEXP_DATA_BASE(const wxChar*) wxEmptyString;
  45 #if wxUSE_UNICODE_UTF8
  46 // FIXME-UTF8: we should have only one wxEmptyString
  47 extern WXDLLIMPEXP_DATA_BASE(const wxStringCharType*) wxEmptyStringImpl;
  48 #endif
  49
  50
  51 // ----------------------------------------------------------------------------
  52 // deal with STL/non-STL/non-STL-but-wxUSE_STD_STRING
  53 // ----------------------------------------------------------------------------
  54
  55 #define wxUSE_STL_BASED_WXSTRING  wxUSE_STL
  56
  57 // in both cases we need to define wxStdString
  58 #if wxUSE_STL_BASED_WXSTRING || wxUSE_STD_STRING
  59
  60 #include "wx/beforestd.h"
  61 #include <string>
  62 #include "wx/afterstd.h"
  63
  64 #ifdef HAVE_STD_WSTRING
  65     typedef std::wstring wxStdWideString;
  66 #else
  67     typedef std::basic_string<wchar_t> wxStdWideString;
  68 #endif
  69
  70 #if wxUSE_UNICODE_WCHAR
  71     typedef wxStdWideString wxStdString;
  72 #else
  73     typedef std::string wxStdString;
  74 #endif
  75
  76 #endif // wxUSE_STL_BASED_WXSTRING || wxUSE_STD_STRING
  77
  78
  79 #if wxUSE_STL_BASED_WXSTRING
  80
  81     // we always want ctor from std::string when using std::string internally
  82     #undef wxUSE_STD_STRING
  83     #define wxUSE_STD_STRING 1
  84
  85     #if (defined(__GNUG__) && (__GNUG__ < 3)) || \
  86         (defined(_MSC_VER) && (_MSC_VER <= 1200))
  87         #define wxSTRING_BASE_HASNT_CLEAR
  88     #endif
  89
  90     typedef wxStdString wxStringImpl;
  91 #else // if !wxUSE_STL_BASED_WXSTRING
  92
  93 // in non-STL mode, compare() is implemented in wxString and not wxStringImpl
  94 #undef HAVE_STD_STRING_COMPARE
  95
  96 // ---------------------------------------------------------------------------
  97 // string data prepended with some housekeeping info (used by wxString class),
  98 // is never used directly (but had to be put here to allow inlining)
  99 // ---------------------------------------------------------------------------
 100
 101 struct WXDLLIMPEXP_BASE wxStringData
 102 {
 103   int     nRefs;        // reference count
 104   size_t  nDataLength,  // actual string length
 105           nAllocLength; // allocated memory size
 106
 107   // mimics declaration 'wxStringCharType data[nAllocLength]'
 108   wxStringCharType* data() const { return (wxStringCharType*)(this + 1); }
 109
 110   // empty string has a special ref count so it's never deleted
 111   bool  IsEmpty()   const { return (nRefs == -1); }
 112   bool  IsShared()  const { return (nRefs > 1);   }
 113
 114   // lock/unlock
 115   void  Lock()   { if ( !IsEmpty() ) nRefs++;                    }
 116
 117   // VC++ will refuse to inline Unlock but profiling shows that it is wrong
 118 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 119   __forceinline
 120 #endif
 121   // VC++ free must take place in same DLL as allocation when using non dll
 122   // run-time library (e.g. Multithreaded instead of Multithreaded DLL)
 123 #if defined(__VISUALC__) && defined(_MT) && !defined(_DLL)
 124   void  Unlock() { if ( !IsEmpty() && --nRefs == 0) Free();  }
 125   // we must not inline deallocation since allocation is not inlined
 126   void  Free();
 127 #else
 128   void  Unlock() { if ( !IsEmpty() && --nRefs == 0) free(this);  }
 129 #endif
 130
 131   // if we had taken control over string memory (GetWriteBuf), it's
 132   // intentionally put in invalid state
 133   void  Validate(bool b)  { nRefs = (b ? 1 : 0); }
 134   bool  IsValid() const   { return (nRefs != 0); }
 135 };
 136
 137 class WXDLLIMPEXP_BASE wxStringImpl
 138 {
 139 public:
 140   // an 'invalid' value for string index, moved to this place due to a CW bug
 141   static const size_t npos;
 142
 143 protected:
 144   // points to data preceded by wxStringData structure with ref count info
 145   wxStringCharType *m_pchData;
 146
 147   // accessor to string data
 148   wxStringData* GetStringData() const { return (wxStringData*)m_pchData - 1; }
 149
 150   // string (re)initialization functions
 151     // initializes the string to the empty value (must be called only from
 152     // ctors, use Reinit() otherwise)
 153 #if wxUSE_UNICODE_UTF8
 154   void Init() { m_pchData = (wxStringCharType *)wxEmptyStringImpl; } // FIXME-UTF8
 155 #else
 156   void Init() { m_pchData = (wxStringCharType *)wxEmptyString; }
 157 #endif
 158     // initializes the string with (a part of) C-string
 159   void InitWith(const wxStringCharType *psz, size_t nPos = 0, size_t nLen = npos);
 160     // as Init, but also frees old data
 161   void Reinit() { GetStringData()->Unlock(); Init(); }
 162
 163   // memory allocation
 164     // allocates memory for string of length nLen
 165   bool AllocBuffer(size_t nLen);
 166     // effectively copies data to string
 167   bool AssignCopy(size_t, const wxStringCharType *);
 168
 169   // append a (sub)string
 170   bool ConcatSelf(size_t nLen, const wxStringCharType *src, size_t nMaxLen);
 171   bool ConcatSelf(size_t nLen, const wxStringCharType *src)
 172     { return ConcatSelf(nLen, src, nLen); }
 173
 174   // functions called before writing to the string: they copy it if there
 175   // are other references to our data (should be the only owner when writing)
 176   bool CopyBeforeWrite();
 177   bool AllocBeforeWrite(size_t);
 178
 179     // compatibility with wxString
 180   bool Alloc(size_t nLen);
 181
 182 public:
 183   // standard types
 184   typedef wxStringCharType value_type;
 185   typedef wxStringCharType char_type;
 186   typedef size_t size_type;
 187   typedef value_type& reference;
 188   typedef const value_type& const_reference;
 189   typedef value_type* pointer;
 190   typedef const value_type* const_pointer;
 191
 192   // macro to define the bulk of iterator and const_iterator classes
 193   #define WX_DEFINE_STRINGIMPL_ITERATOR(iterator_name, ref_type, ptr_type)    \
 194     public:                                                                   \
 195         typedef wxStringCharType value_type;                                  \
 196         typedef ref_type reference;                                           \
 197         typedef ptr_type pointer;                                             \
 198         typedef int difference_type;                                          \
 199                                                                               \
 200         iterator_name(pointer ptr) : m_ptr(ptr) { }                           \
 201                                                                               \
 202         reference operator*() const { return *m_ptr; }                        \
 203                                                                               \
 204         iterator_name& operator++() { m_ptr++; return *this; }                \
 205         iterator_name operator++(int)                                         \
 206         {                                                                     \
 207             const iterator_name tmp(*this);                                   \
 208             m_ptr++;                                                          \
 209             return tmp;                                                       \
 210         }                                                                     \
 211                                                                               \
 212         iterator_name& operator--() { m_ptr--; return *this; }                \
 213         iterator_name operator--(int)                                         \
 214         {                                                                     \
 215             const iterator_name tmp(*this);                                   \
 216             m_ptr--;                                                          \
 217             return tmp;                                                       \
 218         }                                                                     \
 219                                                                               \
 220         iterator_name operator+(int n) const                                  \
 221             { return iterator_name(m_ptr + n); }                              \
 222         iterator_name operator+(size_t n) const                               \
 223             { return iterator_name(m_ptr + n); }                              \
 224         iterator_name operator-(int n) const                                  \
 225             { return iterator_name(m_ptr - n); }                              \
 226         iterator_name operator-(size_t n) const                               \
 227             { return iterator_name(m_ptr - n); }                              \
 228         iterator_name& operator+=(int n)                                      \
 229             { m_ptr += n; return *this; }                                     \
 230         iterator_name& operator+=(size_t n)                                   \
 231             { m_ptr += n; return *this; }                                     \
 232         iterator_name& operator-=(int n)                                      \
 233             { m_ptr -= n; return *this; }                                     \
 234         iterator_name& operator-=(size_t n)                                   \
 235             { m_ptr -= n; return *this; }                                     \
 236                                                                               \
 237         difference_type operator-(const iterator_name& i) const               \
 238             { return m_ptr - i.m_ptr; }                                       \
 239                                                                               \
 240         bool operator==(const iterator_name& i) const                         \
 241           { return m_ptr == i.m_ptr; }                                        \
 242         bool operator!=(const iterator_name& i) const                         \
 243           { return m_ptr != i.m_ptr; }                                        \
 244                                                                               \
 245         bool operator<(const iterator_name& i) const                          \
 246           { return m_ptr < i.m_ptr; }                                         \
 247         bool operator>(const iterator_name& i) const                          \
 248           { return m_ptr > i.m_ptr; }                                         \
 249         bool operator<=(const iterator_name& i) const                         \
 250           { return m_ptr <= i.m_ptr; }                                        \
 251         bool operator>=(const iterator_name& i) const                         \
 252           { return m_ptr >= i.m_ptr; }                                        \
 253                                                                               \
 254     private:                                                                  \
 255         /* for wxStringImpl use only */                                       \
 256         operator pointer() const { return m_ptr; }                            \
 257                                                                               \
 258         friend class WXDLLIMPEXP_BASE wxStringImpl;                           \
 259                                                                               \
 260         pointer m_ptr
 261
 262   // we need to declare const_iterator in wxStringImpl scope, the friend
 263   // declaration inside iterator class itself is not enough, or at least not
 264   // for g++ 3.4 (g++ 4 is ok)
 265   class const_iterator;
 266
 267   class iterator
 268   {
 269     WX_DEFINE_STRINGIMPL_ITERATOR(iterator,
 270                                   wxStringCharType&,
 271                                   wxStringCharType*);
 272
 273     friend class const_iterator;
 274   };
 275
 276   class const_iterator
 277   {
 278   public:
 279       const_iterator(iterator i) : m_ptr(i.m_ptr) { }
 280
 281       WX_DEFINE_STRINGIMPL_ITERATOR(const_iterator,
 282                                     const wxStringCharType&,
 283                                     const wxStringCharType*);
 284   };
 285
 286   #undef WX_DEFINE_STRINGIMPL_ITERATOR
 287
 288
 289   // constructors and destructor
 290     // ctor for an empty string
 291   wxStringImpl() { Init(); }
 292     // copy ctor
 293   wxStringImpl(const wxStringImpl& stringSrc)
 294   {
 295     wxASSERT_MSG( stringSrc.GetStringData()->IsValid(),
 296                   _T("did you forget to call UngetWriteBuf()?") );
 297
 298     if ( stringSrc.empty() ) {
 299       // nothing to do for an empty string
 300       Init();
 301     }
 302     else {
 303       m_pchData = stringSrc.m_pchData;            // share same data
 304       GetStringData()->Lock();                    // => one more copy
 305     }
 306   }
 307     // string containing nRepeat copies of ch
 308   wxStringImpl(size_type nRepeat, wxStringCharType ch);
 309     // ctor takes first nLength characters from C string
 310     // (default value of npos means take all the string)
 311   wxStringImpl(const wxStringCharType *psz)
 312       { InitWith(psz, 0, npos); }
 313   wxStringImpl(const wxStringCharType *psz, size_t nLength)
 314       { InitWith(psz, 0, nLength); }
 315     // take nLen chars starting at nPos
 316   wxStringImpl(const wxStringImpl& str, size_t nPos, size_t nLen)
 317   {
 318     wxASSERT_MSG( str.GetStringData()->IsValid(),
 319                   _T("did you forget to call UngetWriteBuf()?") );
 320     Init();
 321     size_t strLen = str.length() - nPos; nLen = strLen < nLen ? strLen : nLen;
 322     InitWith(str.c_str(), nPos, nLen);
 323   }
 324     // take everything between start and end
 325   wxStringImpl(const_iterator start, const_iterator end);
 326
 327
 328     // ctor from and conversion to std::string
 329 #if wxUSE_STD_STRING
 330   wxStringImpl(const wxStdString& impl)
 331       { InitWith(impl.c_str(), 0, impl.length()); }
 332
 333   operator wxStdString() const
 334       { return wxStdString(c_str(), length()); }
 335 #endif
 336
 337
 338     // dtor is not virtual, this class must not be inherited from!
 339   ~wxStringImpl()
 340   {
 341 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 342       //RN - according to the above VC++ does indeed inline this,
 343       //even though it spits out two warnings
 344       #pragma warning (disable:4714)
 345 #endif
 346
 347       GetStringData()->Unlock();
 348   }
 349
 350 #if defined(__VISUALC__) && (__VISUALC__ >= 1200)
 351     //re-enable inlining warning
 352     #pragma warning (default:4714)
 353 #endif
 354   // overloaded assignment
 355     // from another wxString
 356   wxStringImpl& operator=(const wxStringImpl& stringSrc);
 357     // from a character
 358   wxStringImpl& operator=(wxStringCharType ch);
 359     // from a C string
 360   wxStringImpl& operator=(const wxStringCharType *psz);
 361
 362     // return the length of the string
 363   size_type length() const { return GetStringData()->nDataLength; }
 364     // return the length of the string
 365   size_type size() const { return length(); }
 366     // return the maximum size of the string
 367   size_type max_size() const { return npos; }
 368     // resize the string, filling the space with c if c != 0
 369   void resize(size_t nSize, wxStringCharType ch = '\0');
 370     // delete the contents of the string
 371   void clear() { erase(0, npos); }
 372     // returns true if the string is empty
 373   bool empty() const { return length() == 0; }
 374     // inform string about planned change in size
 375   void reserve(size_t sz) { Alloc(sz); }
 376   size_type capacity() const { return GetStringData()->nAllocLength; }
 377
 378   // lib.string.access
 379     // return the character at position n
 380   value_type at(size_type n) const
 381     { wxASSERT_VALID_INDEX( n ); return m_pchData[n]; }
 382     // returns the writable character at position n
 383   reference at(size_type n)
 384   {
 385     wxASSERT_VALID_INDEX( n );
 386     CopyBeforeWrite();
 387     return m_pchData[n];
 388   } // FIXME-UTF8: not useful for us...?
 389
 390   // lib.string.modifiers
 391     // append elements str[pos], ..., str[pos+n]
 392   wxStringImpl& append(const wxStringImpl& str, size_t pos, size_t n)
 393   {
 394     wxASSERT(pos <= str.length());
 395     ConcatSelf(n, str.c_str() + pos, str.length() - pos);
 396     return *this;
 397   }
 398     // append a string
 399   wxStringImpl& append(const wxStringImpl& str)
 400     { ConcatSelf(str.length(), str.c_str()); return *this; }
 401     // append first n (or all if n == npos) characters of sz
 402   wxStringImpl& append(const wxStringCharType *sz)
 403     { ConcatSelf(Strsize(sz), sz); return *this; }
 404   wxStringImpl& append(const wxStringCharType *sz, size_t n)
 405     { ConcatSelf(n, sz); return *this; }
 406     // append n copies of ch
 407   wxStringImpl& append(size_t n, wxStringCharType ch);
 408     // append from first to last
 409   wxStringImpl& append(const_iterator first, const_iterator last)
 410     { ConcatSelf(last - first, first); return *this; }
 411
 412     // same as `this_string = str'
 413   wxStringImpl& assign(const wxStringImpl& str)
 414     { return *this = str; }
 415     // same as ` = str[pos..pos + n]
 416   wxStringImpl& assign(const wxStringImpl& str, size_t pos, size_t n)
 417     { clear(); return append(str, pos, n); }
 418     // same as `= first n (or all if n == npos) characters of sz'
 419   wxStringImpl& assign(const wxStringCharType *sz)
 420     { clear(); return append(sz, Strsize(sz)); }
 421   wxStringImpl& assign(const wxStringCharType *sz, size_t n)
 422     { clear(); return append(sz, n); }
 423     // same as `= n copies of ch'
 424   wxStringImpl& assign(size_t n, wxStringCharType ch)
 425     { clear(); return append(n, ch); }
 426     // assign from first to last
 427   wxStringImpl& assign(const_iterator first, const_iterator last)
 428     { clear(); return append(first, last); }
 429
 430     // first valid index position
 431   const_iterator begin() const { return m_pchData; }
 432   iterator begin();
 433     // position one after the last valid one
 434   const_iterator end() const { return m_pchData + length(); }
 435   iterator end();
 436
 437     // insert another string
 438   wxStringImpl& insert(size_t nPos, const wxStringImpl& str)
 439   {
 440     wxASSERT( str.GetStringData()->IsValid() );
 441     return insert(nPos, str.c_str(), str.length());
 442   }
 443     // insert n chars of str starting at nStart (in str)
 444   wxStringImpl& insert(size_t nPos, const wxStringImpl& str, size_t nStart, size_t n)
 445   {
 446     wxASSERT( str.GetStringData()->IsValid() );
 447     wxASSERT( nStart < str.length() );
 448     size_t strLen = str.length() - nStart;
 449     n = strLen < n ? strLen : n;
 450     return insert(nPos, str.c_str() + nStart, n);
 451   }
 452     // insert first n (or all if n == npos) characters of sz
 453   wxStringImpl& insert(size_t nPos, const wxStringCharType *sz, size_t n = npos);
 454     // insert n copies of ch
 455   wxStringImpl& insert(size_t nPos, size_t n, wxStringCharType ch)
 456     { return insert(nPos, wxStringImpl(n, ch)); }
 457   iterator insert(iterator it, wxStringCharType ch)
 458     { size_t idx = it - begin(); insert(idx, 1, ch); return begin() + idx; }
 459   void insert(iterator it, const_iterator first, const_iterator last)
 460     { insert(it - begin(), first, last - first); }
 461   void insert(iterator it, size_type n, wxStringCharType ch)
 462     { insert(it - begin(), n, ch); }
 463
 464     // delete characters from nStart to nStart + nLen
 465   wxStringImpl& erase(size_type pos = 0, size_type n = npos);
 466   iterator erase(iterator first, iterator last)
 467   {
 468     size_t idx = first - begin();
 469     erase(idx, last - first);
 470     return begin() + idx;
 471   }
 472   iterator erase(iterator first);
 473
 474   // explicit conversion to C string (use this with printf()!)
 475   const wxStringCharType* c_str() const { return m_pchData; }
 476   const wxStringCharType* data() const { return m_pchData; }
 477
 478     // replaces the substring of length nLen starting at nStart
 479   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringCharType* sz);
 480     // replaces the substring of length nLen starting at nStart
 481   wxStringImpl& replace(size_t nStart, size_t nLen, const wxStringImpl& str)
 482     { return replace(nStart, nLen, str.c_str()); }
 483     // replaces the substring with nCount copies of ch
 484   wxStringImpl& replace(size_t nStart, size_t nLen, size_t nCount, wxStringCharType ch);
 485     // replaces a substring with another substring
 486   wxStringImpl& replace(size_t nStart, size_t nLen,
 487                         const wxStringImpl& str, size_t nStart2, size_t nLen2);
 488     // replaces the substring with first nCount chars of sz
 489   wxStringImpl& replace(size_t nStart, size_t nLen,
 490                         const wxStringCharType* sz, size_t nCount);
 491   wxStringImpl& replace(iterator first, iterator last, const_pointer s)
 492     { return replace(first - begin(), last - first, s); }
 493   wxStringImpl& replace(iterator first, iterator last, const_pointer s,
 494                         size_type n)
 495     { return replace(first - begin(), last - first, s, n); }
 496   wxStringImpl& replace(iterator first, iterator last, const wxStringImpl& s)
 497     { return replace(first - begin(), last - first, s); }
 498   wxStringImpl& replace(iterator first, iterator last, size_type n, wxStringCharType c)
 499     { return replace(first - begin(), last - first, n, c); }
 500   wxStringImpl& replace(iterator first, iterator last,
 501                         const_iterator first1, const_iterator last1)
 502     { return replace(first - begin(), last - first, first1, last1 - first1); }
 503
 504     // swap two strings
 505   void swap(wxStringImpl& str);
 506
 507     // All find() functions take the nStart argument which specifies the
 508     // position to start the search on, the default value is 0. All functions
 509     // return npos if there were no match.
 510
 511     // find a substring
 512   size_t find(const wxStringImpl& str, size_t nStart = 0) const;
 513
 514     // find first n characters of sz
 515   size_t find(const wxStringCharType* sz, size_t nStart = 0, size_t n = npos) const;
 516
 517     // find the first occurrence of character ch after nStart
 518   size_t find(wxStringCharType ch, size_t nStart = 0) const;
 519
 520     // rfind() family is exactly like find() but works right to left
 521
 522     // as find, but from the end
 523   size_t rfind(const wxStringImpl& str, size_t nStart = npos) const;
 524
 525     // as find, but from the end
 526   size_t rfind(const wxStringCharType* sz, size_t nStart = npos,
 527                size_t n = npos) const;
 528     // as find, but from the end
 529   size_t rfind(wxStringCharType ch, size_t nStart = npos) const;
 530
 531   size_type copy(wxStringCharType* s, size_type n, size_type pos = 0);
 532
 533   // substring extraction
 534   wxStringImpl substr(size_t nStart = 0, size_t nLen = npos) const;
 535
 536       // string += string
 537   wxStringImpl& operator+=(const wxStringImpl& s) { return append(s); }
 538       // string += C string
 539   wxStringImpl& operator+=(const wxStringCharType *psz) { return append(psz); }
 540       // string += char
 541   wxStringImpl& operator+=(wxStringCharType ch) { return append(1, ch); }
 542
 543 #if !wxUSE_UNICODE_UTF8
 544   // helpers for wxStringBuffer and wxStringBufferLength
 545   wxStringCharType *DoGetWriteBuf(size_t nLen);
 546   void DoUngetWriteBuf();
 547   void DoUngetWriteBuf(size_t nLen);
 548 #endif
 549
 550 private:
 551 #if wxUSE_UNICODE_UTF8
 552   static size_t Strsize(const wxStringCharType *s) { return strlen(s); }
 553 #else
 554   static size_t Strsize(const wxStringCharType *s) { return wxStrlen(s); }
 555 #endif
 556
 557   friend class WXDLLIMPEXP_BASE wxString;
 558 };
 559
 560 #endif // !wxUSE_STL_BASED_WXSTRING
 561
 562 // don't pollute the library user's name space
 563 #undef wxASSERT_VALID_INDEX
 564
 565 #endif  // _WX_WXSTRINGIMPL_H__